Referred to as SWE02, the new data centre by atNorth will be a 30MW metro site built in close proximity to the company’s first state-of-the-art data centre in Sweden, SWE01.

The data centre will be enabled with heat reuse capabilities and allow excess heat from the data centre to be recovered and recycled for use within the local community, in partnership with Stockholm Exergi.

“We are proud to collaborate with atNorth to recycle heat to meet a local need within Stockholm,” says Peter Sivengård, business project manager, at Stockholm Exergi. “By working together, we can ensure efficient use of our resources and contribute to our country’s circular economy.”

atNorth also said this setup will enable customers to seamlessly deploy connected clusters across both data centre locations.

SWE02 will facilitate seamless cluster operations between both campuses in the area and offer separation of client workloads that bolster redundancy and resilience. The site is expected to go live in Q4 2027.

“The expansion of our presence in Stockholm is a clear reflection of the growing demand for responsible data centre operators who can combine technical excellence with sustainable operations,” said Eyjólfur Magnús Kristinsson, CEO of atNorth.

“As high-performance digital infrastructure becomes increasingly business critical, companies are seeking partners who prioritise energy efficiency, sustainable design and long-term community impact – values that are deeply embedded in atNorth’s growth strategy.”

By expanding in the region, atNorth has helped facilitate a new electrical substation. This will be constructed by network company Ellevio and aim to improve electrical network infrastructure resilience for the surrounding area.

“We are delighted that atNorth has chosen the Kista area for an additional energy efficient data centre campus that will contribute to the city’s climate efforts,” says Karin Wanngård, Mayor of Stockholm. “The development and subsequent operational phase of this data centre will bring employment opportunities for our community and further position Stockholm as a hub for digital infrastructure excellence.”

The news comes shortly after other announcements from atNorth, including the expansion of two data centre campuses in Iceland and the securing of land in Sollefteå, Sweden for a future mega site.

Likewise, atNorth has entered into a heat reuse agreement with Vesforbrænding, Denmark’s largest waste-to-energy company, to repurpose excess heat from its DEN01 data centre campus, alongside the announcement of colocation partnerships with Nokia, Crusoe and 6G AI Sweden AB.

Related stories

How atNorth’s DEN01 is supplying Danish homes with heat

How has the data centre industry evolved in 2025?

Building the Nordics’ digital future

OpenAI is also considering a deal that would allow Amazon to use its AI models,  which run the ChatGPT chatbot, in Amazon’s products and services. 

Amazon employees could also use the models for their work, despite the technology giant announcing job cuts over nearly 16,000 job roles, just days before.

According to reports, if talks are successful, Amazon could supply as much as half of a new funding round worth up to $100 billion. Other possible investors include Nvidia and SoftBank Group, which recently acquired DigitalBridge.

The funding round is not finished and could still change, the anonymous source added.

The Wall Street Journal first reported the $50 billion figure and said Amazon CEO Andy Jassy is leading the talks with Sam Altman, CEO of OpenAI. The report also said OpenAI is trying to raise up to $100 billion from investors.

This follows Amazon being announced as a founding partner alongside Accenture, Barclays, BT, Google, IBM, Intuit, Microsoft, Sage, SAS and Salesforce in the UK’s new AI Skills Boost programme, which aims to upskill 10 million workers by 2030.

However, Amazon’s actions raise questions about whether AI skills will truly safeguard workers, or simply make it easier for companies to reduce headcount.

Andrew McLernon, co-founder and CEO of Interlink, a global, AI-powered business, told Capacity: “AI skills on their own won’t protect jobs, but how companies choose to use AI will. If AI is treated purely as a cost-cutting tool, then yes, it becomes a shortcut to reducing headcount. 

“But when it’s used properly, AI creates headroom. It takes care of the heavy lifting so people can focus on judgement, creativity, experience and building trust, all the things AI simply can’t replicate.”

He continued: “While training millions of workers in AI skills is important, it’s only half the story. AI is now widely accessible, but what truly differentiates organisations is knowing how, when and even when not to use AI. That comes from experience, transparency and people who are confident enough to challenge the data, not just follow it.”

Peter Fedoročko, CTO of GoodData, AI data intelligence platform continued: “Amazon’s decision to cut 16,000 corporate roles is a signal that even the big names are rethinking the realities of automation and post-pandemic scaling. 

“Many companies overexpanded during a period of cheap capital and rapid digital acceleration. Now, they’re confronting what happens when AI, automation, and market efficiency collide with over-hiring,”

He reveals, the takeaway isn’t necessarily cost-cutting – it’s about system design. 

“How do we integrate tech? The companies that succeed will be those that use AI in core workflows early, understand which functions can be automated safely, and reallocate human talent toward higher-order problem-solving.

“AI isn’t eliminating the human factor; it’s redefining it. The value now lies in those who can design, monitor, and collaborate with automated systems – not compete against them.”

According to Fedoročko, the future belongs to organisations that treat AI not as a headcount reduction tool, but as a tool to benefit the workforce. 

“It’s not humans vs machines, it’s about giving the humans the best chance to succeed using a technology that isn’t going away,” he continues.

Trade organisation, GMB’s organiser Rachel Fagan, blasted the technology giant claiming: “Amazon is showing itself for what it is; a company that cannot be trusted to do the right thing by working people in the UK.

“Bosses are overseeing thousands of job losses which will cause huge damage in towns and cities across the country.

“Now is the time for decision makers to recognise Amazon as a company fixated on eye-watering profits at the expense of workers and local people.”

Related stories

Amazon confirms new round of job cuts after internal email leak

Amazon: What we know so far about latest job cuts

Nvidia, Microsoft and Amazon in talks on up to $6bn OpenAI investment

An AI summit held at the House of Lords has called for greater use of agentic AI to “rejuvenate Britain’s sluggish economy”.

The summit, chaired by Centropy PR founder Steven George-Hilley, brought together a range of leaders from across technology, legal, financial and cybersecurity sectors for a discussion on the opportunities and challenges AI presents. Also discussed was the impact of AI technologies on job opportunities and what skills challenges face smaller businesses.

Some of the key themes explored were the role of agentic AI in offering SMEs better access to professional sales and customer management systems, in addition to how critical technology could be in unlocking economic growth.

Agentic AI refers to an autonomous system that is powered by large language models (LLMs) that has the capability to independently plan, make decisions and execute multi-step tasks to achieve specific goals. It moves beyond more passive content generation.

There is plenty of debate over these advanced AI systems and their impact on society and global geopolitics. Debating yesterday, the House of Lords shared views that AI could advance UK society and make it a global business leader – if handled carefully.

“Currently, no method exists to contain or control smarter-than-human AI systems,” Lord Clement-Jones said in Parliament yesterday. “Ministers focus on regulating the use of AI tools rather than their development. But this approach fails fundamentally when facing superintelligence.

“Once a system surpasses human intelligence across all domains, we cannot simply regulate how it is used. We will have lost the ability to control it at all. You cannot regulate the use of something more intelligent than the regulator just sector by sector.”

Some argue for the potential of AI, as the technology continues to work to reshape businesses – improving efficiencies and revealing new opportunities for long-term growth and innovation. The UK’s Autumn 2025 Budget strongly advocated for the upscaling of AI data centres so that the country remains a competitive leader in the technology race.

Likewise, the government’s continued interest in AI Growth Zones is promising for businesses, as demand for AI innovation continues to grow.

Dmitry Tikhomirov, VP, technology solutions at EPAM, argued that seeing tangible value from AI, however, will take more than technology alone.

“It requires business and technology leaders working closely together, with the right platforms and skills in place, so adoption is clearly tied to strategic goals,” he said.

As AI systems rely on large amounts of sensitive data, calls continue for technology businesses and policymakers to collaborate to ensure regulatory compliance and data protection to maintain public trust.

“As network traffic increases, due to the increase in demand for AI ready capacity and the continued growth of high-quality content streaming, the infrastructure powering the internet becomes even more critical,” explained LINX CTO Richard Petrie. “Businesses and end-users expect resilient, low-latency internet connectivity, requiring internet exchange points to facilitate fast data transfer and ensure networks maintain uptime.”

He added: “The demand for high-performance digital services is only going to increase, so it’s important that the UK prioritises investment in infrastructure to support the future of emerging technologies and network growth.”

Summit attendees also warned of the looming “skills cliff edge” in Britian’s workforce, as AI continues to reshape the economy and redefine boardroom priorities. This involved examining the economic impact of AI on job opportunities, as well as sounding alarm over the skills challenges facing smaller businesses.

This comes amid a wave of job cuts impacting the wider technology sector, with AI largely being blamed. Amazon, along with other technology and telecoms giants, have confirmed new job cuts worldwide, as executives re-shift their focus to AI development strategies.

On this, Cisco chairman and chief executive Chuck Robbins suggested to the BBC that some companies could lose their values or fail to recognise the risks of AI technology if they aren’t careful.

“[AI] is going to be bigger than the internet,” he said, arguing that “winners will emerge, the applications and use cases will begin to evolve,” but that some companies “won’t make it” – suggesting that some jobs will be changed or “eliminated” by AI, particularly in areas like customer service where businesses will need fewer people.

Related stories

AI godfather blasts Zuckerberg: Yann LeCun breaks his silence

AMD CEO: Agentic AI marks ‘next chapter of AI‘

Caterpillar to supply 2GW of gas generation for US AI data centre campus

 

Sparkle and Valencia Digital Port Connect (VDPC), the Spanish telecommunications company, have agreed to land the Barracuda subsea cable at Sparkle’s Genoa Landing Platform.

As the first direct Spain-Italy submarine cable, the extension to Barracuda will expand access to major European internet exchange points (IXPs) because of Sparkle’s landing and international connectivity infrastructure.

Working in partnership with private equity firm Teset Capital, Sparkle and VDPC will establish the first direct high-capacity, low-latency submarine route between Spain and Italy. The cable extension hopes to create a 1,070km digital bridge between Valencia and Genoa. Designed with an “open cable system” architecture, Barracuda will feature 12 fibre pairs, each with a capacity of 32 Tbps.

In total, the project has an estimated total investment of €100 million and is scheduled to be completed in three years, with operations expected to begin in 2028.

“This agreement represents an important step in our strategy to position Genoa as a key gateway to Europe,” said Enrico Bagnasco, CEO of Sparkle and chairman of the board at the Global Leadership Forum (GLF). “The landing of Barracuda will strengthen and expand the city’s digital ecosystem, while the additional capacity on the system allows us to further expand our resilient, high-performance Mediterranean network and better serve international connectivity demand”.

Under the terms of the agreement, Barracuda will land at Sparkle’s Genoa Landing Platform to promote resiliency and secure submarine cable landing on the Western Europe coastline. By landing in Genoa, VDPC will gain immediate access to major European hubs, avoiding the cost, time and administrative complexity of deploying a proprietary landing infrastructure.

Sparkle will also acquire infrastructure assets on the Barracuda submarine cable system between Valencia and Genoa and colocation in Valencia Cable Landing Station. This is a fully neutral, scalable infrastructure designed for Barracuda and up to three additional submarine cable systems.

The additional capability hopes to strengthen Sparkle’s connectivity across the Iberian Peninsula, serving both the local and growing West African market – in addition to reinforcing its strategic footprint across the Mediterranean.

Enrique Martín, CEO at VDPC said: “This agreement with Sparkle marks a key milestone in the Barracuda project and confirms that we are advancing in line with our strategic roadmap.”

He added: “Securing Genoa as our landing point and welcoming Sparkle as a long-term customer reinforces Barracuda and Valencia Cable Landing Station as strategic assets for international partners confirming the credibility of. our ambition to have the system fully operational by 2028.”

Related stories

Sparkle launches new quantum-safe framework

euNetworks names Giancarlo Ferro CFO to support next phase of European network growth

Vodafone unveils new subsea cable in Greece

Microsoft suffered its steepest one-day market value loss in years on with shares tumbling over 10% and wiping out approximately $360 billion from its valuation. The rout came despite the company delivering impressive revenue and profit growth, exposing growing friction between Microsoft’s aggressive (AI) strategy and the immediate expectations of investors.

Impressive earnings mask cloud concerns

The tech giant reported robust figures for the quarter ending 31 December, notching up $81.3 billion in revenue (a 17% year-on-year rise) and net income soaring to $38.5 billion. But those headline numbers failed to soothe market nerves. Investors zeroed in on signs that Azure, Microsoft’s powerhouse cloud division and key driver of its AI ambitions, is slowing. Azure’s revenue climbed at a high-30s percentage rate, still strong but falling just short of analyst hopes and hinting at broader challenges in cloud adoption.

Meanwhile, Microsoft’s capital expenditure ballooned to $37.5 billion, up 66% from the previous year, with the lion’s share funnelled into AI infrastructure and sprawling new data centres. The move signalled Microsoft’s commitment to future growth but also fed worries that swelling investment could squeeze near-term profits.

AI bets: Opportunity or overexposure?

Central to Microsoft’s AI charge is its high-profile partnership with OpenAI, the creator of ChatGPT. Nearly half of the company’s remaining revenue commitments are now linked to OpenAI-related projects, underscoring both the immense potential and the looming risks of such a concentrated play. Analysts warn that any shifts in OpenAI’s fortunes could have a material impact on Microsoft’s own outlook.

Addressing the sell-off during the company’s earnings call, CEO Satya Nadella doubled down on Microsoft’s long-term approach. He painted the surging AI spend as foundational to the company’s future, not a drag on short-term performance.

Nadella, spoke out at the World Economic Forum in Davos earlier this month hitting back at claims of an “AI bubble”, arguing that the technology’s true power lies not in hype or speculation, but in tangible improvements to productivity, industry practice, and societal outcomes.

Addressing a packed audience as part of WEF’s spotlight on tech innovation, Nadella called for a shift in perspective. “So many people talk about there may be an AI bubble,” he remarked, “but the democratisation and diffusion of technology are what really transform demand. The ultimate winners will be those who adopt and apply AI fastest -not just the creators of the technology.”

Market backlash

Despite these reassurances, the market’s reaction was swift and punishing. The sharp drop in Microsoft’s share price reflected a broader investor unease around ballooning capital outlays – particularly for AI.

Microsoft’s leadership insists they are playing the long game. CFO Amy Hood pointed to ongoing investments in data centres and AI infrastructure as laying the groundwork for future efficiency and margin improvement.

Microsoft is under increasing scrutiny to deliver tangible, near-term financial returns even as it invests billions for the future. The heavy reliance on OpenAI projects raises questions about concentration risk, while slowing cloud momentum and record capital expenditures suggest that the path to AI-powered profitability could test investor patience.

RELATED STORIES

Nvidia, Microsoft and Amazon in talks on up to $6bn OpenAI investment

Microsoft CEO hits back at AI bubble claims

Located in Kronstorf in the Linz-Land district of Austria, the data centre will be neatly between Vienna and Salzburg, as Google steps up its plans in the country.

Google has not yet commented publicly on the news, nor shared how much it is investing, with the company expected to communicate what’s upcoming after the official review process is complete.

“The building law procedure for Google’s project has been completed and has been approved,” shared Christian Kolarik, Mayor of Kronstorf, as reported by Kronen Zeitung.

He also confirmed: “Google has submitted a concrete project to both the trade authority – the Linz-Land district – and the building authority – the municipality of Kronstorf.”

Google’s data centre timeline for Austria has been slow moving, with the technology giant first purchasing a 70-hectare plot back in 2008. The company is now reportedly moving ahead with plans, with the upcoming data centre marking the first company-owned facility Google has in Austria.

“[Google has] submitted their planning applications and could have permits in place by the end of the year,” said Anthony Scott-Conaty, senior consultant, data centres, EMEA & US at JDL Recruitment. “This is a great example of why getting the right site in Europe can be anything but straightforward.”

Kronen Zeitung reported construction of the data centre could commence soon, with a main building consisting of roughly 29,000 square metres, where computing halls and servers are to be stationed.

The proposed location of the data centre will be close enough to engage talent but also offers enough space for the needs of a hyperscale data centre.

Austria’s data centre market can be complex, as permitting and planning can often be slow. However, it is currently experiencing strong growth and was predicted to surge to US$250.12bn in 2026 to $352.64 billion by 2031, according to Mordor Intelligence. This movement is currently being driven by AI demand, cloud adoption and a high reliance on renewable energy.

“The market is heating up. Austrian data centre capacity is set to grow, and sites like this are becoming increasingly rare,” Scott-Conaty explained. “For those delivering data centres in Europe, it’s a great example of how persistence and strategic planning can pay off… even if it takes nearly two decades!”

This new development comes as other major tech companies are also expanding their cloud infrastructure in Austria. For instance, Microsoft launched a cloud region in Austria in July 2025 to offer businesses access to advanced cloud and AI technologies.

Consisting of three data centres, Microsoft said at the time its facilities will be covered by 100% renewable energy. It also set a goal of training 300,000 people in Austria in digital skills by the end of 2025.

Related stories

Google DeepMind to build first automated research lab in UK

Microsoft CEO hits back at AI bubble claims

KKR and Oak Hill back Global Technical Realty with nearly $2bn to fuel European data centre expansion

 

The campus is a large-scale power and infrastructure project designed to support hyperscale and enterprise AI data centres.

Under the agreement, Caterpillar will deliver Cat G3516 Fast Response natural gas generator sets, with on-site support provided by dealer Boyd CAT.

The first phase of the project will see 2GW brought online in 2027, with equipment deliveries scheduled between September 2026 and August 2027. Further phased expansion is planned over time, with total capacity expected to reach up to 8GW.

Located on a 2,380-acre site near Point Pleasant, West Virginia, the Monarch Compute Campus is designed as a fully self-supplied, behind-the-meter power platform, generating electricity onsite without the need for incremental utility transmission or distribution infrastructure.

The project holds an existing microgrid designation from the State of West Virginia, allowing new capacity to be deployed on an accelerated timeline.

The gas generation plant will be augmented with battery energy storage systems, enabling it to manage the extreme and rapid load swings associated with AI workloads.

Caterpillar’s G3516 units can ramp from zero to full load in approximately seven seconds, making them well suited to continuous-duty data centre applications requiring high power quality and fast response.

The generators will operate on natural gas and incorporate advanced emissions controls, including selective catalytic reduction (SCR), to meet air-permitting requirements and support ultra-low emissions performance.

According to the companies, the Monarch platform is designed to have no impact on existing utility customers, with onsite generation intended to support grid reliability and resilience without increasing rates or adding costs elsewhere. Advanced monitoring and control systems will be used to ensure high availability and predictable lifecycle performance.

Alongside the equipment order, Caterpillar and AIP Corp have also entered into a strategic alliance agreement, covering phased expansion planning, operations and maintenance strategy, service readiness and long-term lifecycle support. The agreement also includes vendor equipment financing through Caterpillar Financial, aligned with delivery phasing.

RELATED STORIES

Caterpillar’s hybrid vision: Powering the world’s mobile future

AI data centre boom strains power grids and cooling supply chains: report

His comments come as businesses worldwide are embracing AI and integrating it into their strategic goals. However, Cisco chairman and chief executive Chuck Robbins suggested that some companies could lose their values or fail to recognise the risks of AI technology if they aren’t careful.

Robbins told the BBC yesterday that AI will be “bigger than the Internet,” but that some companies “won’t make it”. He suggested that some jobs will be changed or “eliminated” by AI, particularly in areas like customer service where businesses will need fewer people.

Urging workers to embrace AI, rather than fear it, Robbins said: “You shouldn’t worry as much about AI taking your job as you should worry about someone who’s very good using AI taking your job.”

His comments come as job cuts are happening worldwide, particularly in larger technology companies, as executives re-shift their focus to strategies like AI development or cost-cutting.

Ericsson, Meta and TCS have already announced workforce reduction in 2026. Yesterday, after a draft internal message from a senior executive was mistakenly shared with employees, Amazon confirmed it was making a new wave of global job cuts.

Many of these cuts have been attributed to AI, as businesses seek to invest further, which sparked wider concerns over the technology ‘taking jobs’ away from workers. Amazon first revealed its plans to eliminate 30,000 jobs in October, claiming at the time that the cuts were largely driven by advances in AI.

AI investment is roaring ahead, with some claiming that the bubble is set to burst. Robbins has likened the movement to the dotcom boom and subsequent crash in the late 1990s, shaking global markets and leading to some companies declaring bankruptcy.

Capacity recently reported that the strategic logic appears to be companies positioning themselves to dominate the next wave of AI-driven services, from enterprise productivity tools to generative content platforms.

However, industry leaders like Microsoft CEO Satya Nadella have challenged the ‘AI bubble’ notion, arguing that the technology’s true power lies not in hype or speculation, but in actual positive outcomes to productivity and social impact.

“So many people talk about there may be an AI bubble,” Nadella said at Davos, “but the democratisation and diffusion of technology are what really transform demand. The ultimate winners will be those who adopt and apply AI fastest – not just the creators of the technology.”

As far as Cisco is concerned, as one of the world’s leading technology companies, the company is responsible for some of the most critical IT infrastructure that enables daily AI use. After seeing an 80% value drop after the dotcom crash, the company has successfully rebuilt itself and partners with global AI and chip leaders like Nvidia to facilitate the AI infrastructure boom.

“There’s been a lot of discussion about: ‘Is this a bubble?’. And the answer is probably yes, but we had a bubble in 2000 with the internet. And look at where we are today,” Robbins said.

“So the winners emerge, and there’s carnage along the way, but [AI] is going to be bigger than the internet,” he said. “The winners will emerge, the applications and use cases will begin to evolve.”

Robbins added that AI will make “lots of things better,” but also has “potential risks we all have to mitigate”.

In addition to driving AI forward, Robbins is also chair of the Business Roundtable, where he represents the technology business community to US President Donald Trump.

President Trump has been very vocal about AI, even before he took office again, leaning into data centre investment across the US and launching the country’s AI Action Plan last year to preserve the nation’s leadership in AI technology.

When the BBC asked Robbins about speaking with the president, he said it’s “probably the most accessible administration that we’ve had in decades. So they’re very open. We have lots of dialogue.

“We don’t always agree, but we at least have the dialogue.”

Related stories

Cisco: Unlocking opportunities for AI and quantum in 2026

Cisco, Asiacell to boost network reliability across Iraq

AMD, Cisco, HUMAIN: Partnering to advance AI infrastructure

The list celebrates the visionaries, strategists, and decision-makers who are shaping the Middle East’s digital future, spanning telecom, cloud, data centres, subsea networks, and technology entrepreneurship.

Each individual has demonstrated exceptional impact, from pioneering transformative projects to enabling national digital agendas and accelerating technological adoption.

“This year’s Power 50 highlights the remarkable talent steering the Middle East’s digital transformation,” said Vanessa Barbe, managing director at Capacity. “These leaders are not only building the infrastructure of today but also shaping the strategies and innovations that will define connectivity, resilience, and growth for years to come.”

The Power 50 also reflects the broader momentum of the Middle East’s digital sector. Investment in subsea cable systems, hyperscale data centres, fibre networks, and cloud platforms is surging, with governments and private players actively advancing national connectivity and innovation strategies. The winners exemplify the individuals turning vision into reality, ensuring the region remains at the forefront of digital infrastructure and technological leadership.

Capacity will formally unveil the full Power 50 in Capacity magazine at Capacity Middle East 2026, offering an opportunity for industry leaders to celebrate the achievements of their peers and foster collaboration across the sector.

For a full round up of all Middle East news click here.

Capacity Middle East is the region’s leading digital infrastructure event, uniting over 3,500 executives from more than 90 countries for visionary content and unrivalled networking and business opportunities. Capacity Middle East unites the entire ecosystem to drive digital transformation, unlock the value of AI, and enhance connectivity for all. With a reputation for attracting the leading regional and international carriers,

The event continues to attract increased attendance from data centre operators and hyperscalers through Datacloud Middle East. From fibre operators, international carriers, towers, data centres and hyperscalers to investment, content providers, vendors and construction, Capacity Middle East is the must attend event for anyone doing business in the region. For more information click here.

Capacity Power 50: Middle East

 

If completed, it would be one of the largest private funding rounds in tech history, underscoring how fiercely companies are competing to shape the future of AI.

Reports suggest Nvidia could contribute as much as $30 billion, building on its long-standing role as a supplier of the specialised chips that power OpenAI’s AI models.

Microsoft, a strategic partner whose cloud and software products already integrate OpenAI technology, may invest less than $10 billion, while Amazon, a potential new backer, could commit more than $10 billion, possibly exceeding $20 billion.

Together, these discussions could total up to $60 billion, though terms are still being finalised and no official confirmations have been made.

The talks highlight OpenAI’s rapid growth and the soaring costs of running cutting-edge AI models, particularly around computing and infrastructure. Fresh capital from major tech players would help sustain its expansion while keeping the company competitive in a market where speed and scale matter.

Amazon’s potential involvement also signals a possible shift in the cloud landscape. OpenAI has historically relied on Microsoft’s Azure platform, but additional investment from Amazon could open the door to broader infrastructure and commercial collaboration with AWS.

RELATED STORIES

DeepSeek in spotlight: Nvidia’s China AI support draws fresh US scrutiny

Nvidia throws CoreWeave £2bn lifeline, Huang blasts critics

It has been reported this week that Intel shares are down 13% this week, after the semiconductor company has struggled to meet the demands of rapidly growing interest in AI data centres.

The company forecast current-quarter revenue between US$11.7 billion and $12.7 billion, compared with analysts’ average estimate of $12.51 billion, according to LSEG data.

It had been hoped that speedy data centre buildouts by technology giants would drive sales for Intel’s server chips, given booming interest in AI. However, Intel has had some difficulties in predicting global chip markets and has fallen behind fellow chipmakers Nvidia and AMD.

The company’s finance chief David Zinsner told Reuters during an interview that the company was caught off guard by surging demand for the sever central processors that support AI chips.

Despite powering full-steam ahead and seeing shares rise by 40% in the past month, the company’s data centre dealmaking has slowed. Zinsner reportedly said that, despite owning its own factories, Intel has a lag time in changing the types of chips it makes and didn’t expect data centre demand to change.

“In the short term, I’m disappointed that we are not able to fully meet the demand in our markets,” Intel chief executive Lip-Bu Tan told analysts during a conference call, adding that the company was “working aggressively to grow supply to meet strong customer demand”.

Intel recently launched its new personal computer chip, which Tan referred as an “important milestone” for the company, as it hopes to lead again in personal computers. The company had been eyeing a memory comeback last year, but more recent news about the global shortage in memory chips could thwart these plans, as the shortage is expected to impact sales across the industry and undersupply AI.

AI demand continues worldwide, having sent cloud computing giants scrambling to upgrade existing infrastructure and develop new partnerships to stay ahead of the technology race.

Chips have even become a political conversation, as the US and China continue to debate over exports – with the US expanding export curbs on advanced chips and chipmaking tools at the end of last year.

Nvidia remains one of the global powerhouses and posted a revenue of $57 billion in the three months ending October 2025, well-surpassing analyst estimates.

Moving forward, Tan has implemented a turnaround strategy focused on cutting costs and developing a fresh product road map. Zinsner also said to Reuters that Intel has held off on investing heavily in its next-generation manufacturing process (14A) while waiting for a large customer.

Intel’s stock did gain 84% in 2025, after more than a 60% drop in its share price in 2024.

Related stories

Nvidia follows Trump in acquiring stake in Intel

HUMAIN, Infra agree $1.2bn framework to boost Saudi AI data centres

Motivair CEO: Power and cooling tech ‘absolutely critical’ to enabling AI

Ericsson reports fourth quarter results and full-year results 2025, finding Q4 saw 12% growth in segment cloud software and services.

Sales increased for the telco company by 6% year-on-year in the fourth quarter, with markets across Europe, the Middle East and Africa growing, in addition to South East Asia and India.

The company added it had benefitted from cost cuts and the sale of its iconectiv business.

Such growth led to Ericsson launching a share buyback this morning, as the company’s cash position sharply improved. The Board will propose an increased dividend of SEK 3.00 per share, and will also seek a mandate for a share buyback of SEK 15 b, Ericsson said.

“Our Q4 results demonstrate solid execution of our strategy priorities. It is encouraging that we delivered organic growth in a flattish RAN market environment through our efforts in mission critical networks, 5G core and Enterprise,” said Börje Ekholm, President and CEO at Ericsson.

Operational actions Ericsson has taken in recent years have now led to improved margins and cash flow for the company, with a ninth consecutive quarter of year-over-year adjusted EBITA margin expansion.

R&D investments to extend technology leadership have continued the company’s success, it said, as it keeps focus on AI-native, secure and autonomous networks. Moving further into 2026, Ekholm said Ericsson expects the RAN market to be flat, but that mission critical and enterprise markets where the company is well-positioned are expected to grow.

“In this environment, we plan to increase investments in defence during 2026 while continuing to optimise our cost base to support margins and cash flow generation,” he said.

It is expected that Ericsson could regain ground in Europe after the European Commission proposed phasing out high-risk suppliers in critical business sectors.

In order to maintain growth amid US import tariffs being introduced, Ericsson moved quickly to restructure to support its weaker 5G investments. In Central and Eastern Europe, in addition to Latin America, the company expects 5G subscription penetration to grow by more than 50% up to 2031.

The company also said at the start of the year it would be cutting 1,600 jobs in Sweden to protect its competitive position.

Related stories

Ericsson: Turbocharging 5G, market collaboration and accelerating UK connectivity

Vodafone CEO: ‘Europe is lagging in 5G rollout’

How can the UK help European telecoms business in 2025?

The initiative will triple capacity across around 1,250 kilometres of fibre routes linking key cities including Stockholm, Gävle, Gothenburg, Oslo and Helsinki.

The project, which began in late 2025 and will run through to 2027, will see new high-capacity fibre cables added to existing infrastructure across Finland, Sweden, Norway and Denmark.

GlobalConnect said the upgrade is designed to futureproof critical digital corridors as demand accelerates, driven largely by the rapid growth of data centres and AI-related workloads.

“Much of the digital infrastructure in the Nordics is more than 20 years old and is running out of capacity,” said Pär Jansson, SVP, GlobalConnect Carrier. “We are now futureproofing it to keep pace with accelerating demand, primarily driven by the rapid growth of data centres and AI hotspots across the region.”

The programme represents an estimated investment of around €40 million and is expected to increase capacity by 300% on reinforced routes. Construction is already underway at multiple land-based sites in Norway and Sweden, with further work planned across Finland from early 2026.

Alongside the terrestrial upgrades, GlobalConnect is also expanding its subsea footprint. The company has announced multiple new subsea cable projects linking Sweden with Finland, Åland, Estonia and Gotland, covering approximately 240 kilometres. These routes are intended to create additional digital highways across the Baltic Sea and improve redundancy across the regional network.

“This project represents one of the most extensive upgrades to the Nordic internet and marks an important milestone for GlobalConnect in today’s geopolitical climate,” Jansson said. “By tripling network capacity and adding several new subsea cables, we are strengthening regional redundancy and resilience for decades to come.”

The backbone upgrade forms part of GlobalConnect’s broader Bifrost programme, a long-term initiative launched in 2023 to modernise the Nordic fibre network with more than 3,000 kilometres of new infrastructure.

Once complete, the expanded network is expected to support the region’s connectivity needs well into the future.

RELATED STORIES

Telenor expands fibre reach with $597m GlobalConnect deal

GlobalConnect completes Nordic super cable project

As a result, the agreement supports the development of AI and digital infrastructure projects across Saudi Arabia.

According to HUMAIN, the Framework Agreement sets out non-binding financing terms for the development of up to 250 MW of hyperscale AI data centre capacity.

The data centres will use GPUs for AI training and inference and will serve HUMAIN’s local, regional and global customers.

HUMAIN and Infra have also agreed to explore setting up an AI data centre investment platform, which would be backed by both organisations and designed to attract global and local institutional investors to support the expansion of HUMAIN’s AI strategy.

HUMAIN CEO, Tareq Amin, said: “Demand growth for advanced compute is intensifying, and this Framework Agreement positions HUMAIN to respond with speed and scale. In partnership with Infra, our goal is to deliver world-class AI data centre infrastructure that enterprises can rely on as their compute needs grow more complex.”

Infra CEO, Eng. Esmail Alsallom, continued: “Today’s Framework Agreement is an important step in expanding Infra’s role of unlocking infrastructure investment opportunities in the Kingdom.

“Our partnership with HUMAIN will activate new pathways to grow institutional investment and develop the digital economy through enabling AI infrastructure.”

RELATED STORIES

Data centre of the month: XDS Datacentres 10MW Saudi Arabia project

AMD, Cisco, HUMAIN: Partnering to advance AI infrastructure

stc partners with Humain, center3 to develop and operate data centres

 

Last year, the telecoms giant revealed revenues climbed up to £170 million, representing a 25% year-on-year increase, while adjusted EBITDA surged 460% to £29 million, driven by accelerating adoption across the network.

According to the telecoms giant, it has achieved over 20% penetration in its consumer footprint, with more than 70% of households switching broadband providers opting for the CityFibre network where available.

As a result, the company is “on track” to exceed 30% penetration by the end of this year.

Alongside this, the CityFibre revealed strategic priorities for the brand, which included expanding its wholesale-only network to eight million UK premises, driving take-up across all market vertices and delivery market-leading products and services across the UK’s most advanced fill fibre.

Simon Holden, CEO of CityFibre, said: “CityFibre’s wholesale business model has always set us apart and, as our network fills up with customers, we have the momentum to bring about a long-term, competitive alternative to BT Openreach and deliver significant benefits for the UK.”

But why is the UK fibre market attracting so much attention from companies right now?

James Robinson, senior analyst at Assembly Research, points to a “confluence of factors, including macroeconomic, financial and commercial ones.” 

This comes as rising demand for high-speed broadband, regulatory support for network competition and the need to achieve scale are all driving consolidation and investment across the industry.

This consolidation is already visible across the UK fibre market, with the types of companies involved in this activity varying.

CityFibre itself has been active, it acquired Lit Fibre, adding up to 300,000 premises to its portfolio, as part of its drive toward an eight‑million‑premises network. 

It also acquired Connexin’s fibre network, extending coverage in Hull, the East Riding and beyond while taking on its Project Gigabit contract. 

Alongside this, earlier deals include CityFibre’s purchase of FibreNation from TalkTalk in 2020.

Robinson added: “Everything we’ve seen so far has been fairly small-scale, involving players operating at the wholesale level only or as vertically integrated outfits.”

So far, the market hasn’t shown major regulatory resistance to consolidation.

Robinson continued: “The majority of transactions would not be expected to pose risks for competition or consumers,” and there are no significant M&A hurdles currently in play unless a major company like Openreach enters the acquisitions market.

He also revealed, operators are also competing on both price and service, claiming some of the strategies companies use to compete effectively in the UK market include “can hinge on price and customer service, or both. Certain altnets have materially discounted retail prices, allegedly to try to prop up short-term penetration.”

RELATED STORIES

CityFibre secures £2.3bn investment to power next phase of UK full fibre rollout

Sky and CityFibre boost UK broadband speeds

Leading sustainable data centre operator Applied Digital has broken ground on Delta Forge 1, an AI Factory campus.

Delta Forge 1 is designed to initially support 430MW of total utility power across two buildings, enabling up to 300MW of critical IT load. It could also be scaled considerably in 2028 and beyond, translating available power into live, high density AI capacity for hyperscale customers.

As a purpose-built campus, Delta Forge 1 has been engineered to support the power integration, advanced cooling and the operations required to run large-scale AI workloads reliably and efficiently.

“Delta Forge 1 represents the next stage of Applied Digital’s growth, as we continue to deliver AI infrastructure through disciplined execution,” said Wes Cummins, chairman and CEO of Applied Digital. “AI Factories succeed or fail based on how effectively power, cooling and operations are integrated.”

He added: “We believe this campus will be built to scale alongside hyperscale demand while delivering operational certainty for customers and lasting value for the communities where we operate.”

As demand for AI soars, converting power into operational AI capacity on schedule and at scale is a key challenge for data centre operators. Applied Digital said it aims to address this challenge by building AI Factories designed for what it calls “repeatable execution” across a growing portfolio of campuses to ensure consistency as scale increases.

The company broke ground on its US$3 billion AI Factory in North Dakota in September 2025 as it sought to expand the state’s role as an AI infrastructure leader in the US. Referred to as Polaris Forge 2, the site aimed to build on the success of its Polaris Forge 1 Ellendale campus – which was featured as Capacity’s data centre of the month in October 2025.

At the time, Cummins told Capacity that Applied Digital’s priority was to scale.

““We’re integrating advanced cooling technologies and modular power systems to optimise performance and sustainability as demand intensifies,” he said. “Partnering closely with CoreWeave will provide seamless deployment of AI workloads that require massive compute resources and flexibility.”

He added: “Ellendale won’t just keep pace with the AI revolution; it will help define the future of high-performance, sustainable AI infrastructure, all while fostering deep, long-term partnerships with the local community.”

For Delta Forge 1, the campus will initially consist of two 150MW facilities spanning more than 500 acres. Once operational – which is expected in mid-2027 – the campus could support more than 200 full-time employees and additional long-term contractors.

The company is advancing Delta Forge 1 in discussions with another prospective hyperscale customer and is eager to showcase the strong demand for purpose-built AI infrastructure that can support high-density workloads at scale.

“We expect Delta Forge 1 to leverage Applied Digital’s proprietary AI Factory blueprint, refined through the company’s Polaris Forge campuses in North Dakota and designed to be replicated across markets with consistency and precision,” Applied Digital said in its press release.

“We believe this repeatable platform approach enables Applied Digital to deliver multiple hyperscale-ready campuses without sacrificing performance, reliability, or responsible development.”

Related stories

Applied Digital announces US$5bn AI Factory lease at North Dakota campus

Data centres could rely less on chillers, but heat rejection remains essential

Motivair CEO: Power and cooling tech & ‘absolutely critical’ to enabling AI

SpaceX is reported to be laying the groundwork for what could become one of the largest initial public offerings in financial history. According to recent reports Elon Musk’s privately held aerospace company has lined up four major Wall Street banks to advise on a possible public listing as early as this year.

While SpaceX has not confirmed its plans, the banks involved are reported to include Morgan Stanley, Goldman Sachs, JPMorgan Chase and Bank of America, with Morgan Stanley widely viewed as the leading contender due to its long-standing relationship with Musk.

Discussions are still at an early stage, but nevertheless, the fact that SpaceX is engaging with multiple investment banks marks its clearest signal yet that an eventual listing is no longer hypothetical.

SpaceX’s valuation has risen dramatically in recent years, fuelled by a combination of technical dominance, aggressive launch economics and the rapid expansion of its Starlink satellite internet business.

Private share sales in late 2024 valued the company at roughly $180 billion, already making it one of the most valuable private companies in the world. Some analysts believe a public listing could push that figure substantially higher, potentially rivalling the largest IPOs ever completed.

Over the past year, the company has continued to break its own records for orbital launches, driven primarily by Falcon 9 missions deploying Starlink satellites. SpaceX now accounts for the overwhelming majority of U.S. orbital launches, and its rapid cadence has fundamentally altered global launch economics.

The satellite broadband service now operates thousands of satellites in low Earth orbit and serves millions of customers worldwide, including consumers, businesses, maritime operators and government agencies.

Recent reporting suggests Starlink is generating several billion dollars in annual revenue, with margins improving as the constellation scales. For public market investors, Starlink provides a recurring revenue stream that complements SpaceX’s historically cyclical launch business.

Alongside Starlink’s growth, SpaceX has continued to push forward with its most ambitious program: Starship. The fully reusable super-heavy launch system is intended to carry massive payloads to orbit, support NASA’s Artemis lunar missions and eventually enable human missions to Mars. While Starship remains in a developmental phase, test flights in 2024 and 2025 demonstrated major technical progress, including controlled re-entries and increasingly complex flight profiles. Regulatory approvals have also expanded, with U.S. authorities allowing a higher annual launch cadence from SpaceX’s Texas facilities.

Starship, if successfully commercialised, could open entirely new markets in satellite deployment, space infrastructure and deep-space missions.

Musk has previously expressed scepticism about public markets, arguing that short-term investor pressures can conflict with innovation-driven engineering timelines. That tension is likely to resurface if SpaceX proceeds with a listing, particularly given the capital-intensive and technically complex nature of its projects.

RELATED STORIES

xAI raises $20bn in Series E funding amid Grok controversy

Satellite shake-out arrives early as Starlink tightens grip on NTN market: report

If chips are engineered in a way to not require cooling, the data centre market evolved to be more zero-water focused. However, heat rejection is still critical to the industry, given continued demands for AI technology.

Nvidia CEO Jensen Huang made comments at the start of the month, suggesting that the company’s next-generation Rubin chips can be cooled using water at temperatures that do not require traditional water chillers. This sparked concern in the data centre cooling market, causing several companies to witness their stocks fall.

In the wake of such comments, Capacity spoke with several data centre executives about what a zero-water industry could look like and how cooling markets could evolve in 2026.

What zero water data centres could look like

Reporting on Nvidia’s Rubin platform suggests racks can run with 45°C water, which could ultimately reduce reliance on the more traditional water chillers for many data centre designs. Indeed, Huang’s comments sparked Trane Technologies and Johnson Controls’ stocks to drop suddenly.

Ozgur Duzgunoglu, design and engineering director at Telehouse, explained that, while these designs can lower energy use and simplify cooling, high-density compute still needs a reliable path to reject heat in peak conditions.

“Zero or near zero on site water use becomes a design outcome, using dry heat rejection, air economisers and sealed loops that avoid evaporation, with liquid cooling supporting dense AI,” he said. “Next comes wider adoption, more testing of liquid cooling options and more heat reuse, supported by transparent tracking of PUE and WUE.”

The ‘hot-water cooling’ concept is a novel idea, given that data centres currently use a significant amount of power for chiller systems and provides refrigerated coolant to the servers. Hot-water cooling capability of a server can be enabled by advanced cold plates, which reduce thermal resistance between the chip and coolant, and intelligent CDUs, which coordinate the coolant flow with the power infrastructure – leading to greater efficiencies and better reliability.

“Hot-water cooling will reduce the energy needed for chiller systems greatly and enable reallocation of this energy to more AI systems, cooling distribution units (CDUs) and electrical infrastructure,” explained Dr. Peter de Bock, vice president, data center energy and cooling technology at Eaton.  This is an exciting time for innovation in data centre design and operation. This approach is estimated to have the potential for up to 33% more AI factory output per grid connection, substantially more efficient.”

Using less chillers could be a positive step in a more sustainable direction for power and water, particularly as the industry seeks to move towards warm-water direct liquid cooling. However, it doesn’t mean ‘no cooling’ or ‘zero water’ altogether.

“Heat must still be rejected and the local climate will govern whether systems can operate year-round without evaporative backup,” said Bruno Berti, senior vice president of global product management at NTT Global Data Centers.

Aligning ‘economics with sustainability’

The AI boom has inevitably prompted a new era of liquid cooling for data centres, enabling power and cooling strategies across the industry.

Speaking with Capacity earlier in the week, Motivair by Schneider Electric CEO Rich Whitmore explained their solutions are designed to keep pace with the chip and silicon evolution to be able to deliver next-gen performance.

“Data centre success now hinges on delivering scalable, reliable, efficient infrastructure solutions that match the next generation of AI Factory deployments,” he said, telling Capacity exclusively that Motivair is working “in the background to help enable these chips and technology.”

He added: “The current conversation of power and cooling is absolutely critical to enable AI infrastructure worldwide. We’re meeting that moment with proven liquid cooling solutions that scale with our customers’ needs.”

With AI continuing to shape the industry, data centre companies could find new opportunities in advancing efficiency – not just at the chip, but also at the critical power and cooling infrastructure level.

“The most exciting part about leaps in efficiency improvements is that it aligns economics with sustainability. It means that AI factories can do more with the power they have,” de Bock added.

Berti said: “The trajectory is clear: higher-temperature liquid cooling and smarter heat rejection will keep pushing facilities toward low- or no-water operation where conditions allow.

“As an industry, we must replace rumour with rigour. It’s critical that we continue to educate communities about water use – what is actually used, what is re-used and how we work to protect shared resources.”

 

Some executives featured in this article have contributed to a data centre feature in an upcoming Capacity Magazine edition. The edition will celebrate the 25th anniversary of our MetroConnect USA event. Find out more information below.

 

Related stories

AI data centre boom strains power grids and cooling supply chains: report

EMEA key to unlocking data centre potential, Vertiv EMEA president says

Marc Ganzi: Why power & AI will redefine the data centre industry

Tecto Data Centers has invested approximately R$200 million (~US$38-39 million) to develop a new data centre in Porto Alegre, Rio Grande do Sul.

Referred to as TPOA1, the facility will be developed near Tecto’s own 33,000-square-metre plot of land located in the Sarandi neighbourhood. The data centre is expected to have a total power capacity of 20 megawatts (MW) and will be connected to the Malbec submarine cable, which links Brazil and Argentina – connecting Rio de Janeiro, São Paulo, Buenos Aires and soon Porto Alegre, currently under construction.

“This project reinforces Tecto’s strategy to expand its operations and infrastructure in key regions of the country, combining resilient connectivity, energy efficiency and proximity to major data consumption hubs,” says Tito Costa, CRO of Tecto.

The news came the same day as V.tal announced it would be building a new submarine cable to connect Brazil to the US. Linking the US and Brazil and hoping to strengthen connectivity between both countries, the cable project will support growing AI demand and deliver high-capacity data traffic connectivity.

Connecting to V.tal’s Malbec cable, TPOA1 will gain access to significant infrastructure and connectivity all over South America. The Malbec cable spans 2,500km already and is gaining a 280km branch unit scheduled for 2027. The network interconnects Rio de Janeiro with Fortaleza, Venezuela, Colombia, the US and Bermuda through V.tal’s entire 26,000 km submarine cable infrastructure.

Tecto’s other data centres in Fortaleza, Ceará, Barranquilla and Colombia will also be connected.

Costa added: “The direct connection to V.tal’s Malbec cable positions Porto Alegre as a strategic point for critical applications, cloud services and content delivery.”

In its first phase, TPOA1 is scheduled to begin operations in the fourth quarter of 2026, with 3MW of capacity to be delivered. Expansion will be gradual, with the project being carried out through retrofit of an existing warehouse on the site, which is being adapted to suit the needs of high-performance data centres.

Notably, the data centre is expected to feature a closed-loop air-cooling system with zero water consumption – something that is being discussed plenty in the data centre industry currently. The facility will also operate with 100% renewable energy.

“The use of solar energy and efficient cooling solutions is part of our commitment to growing responsibly and innovatively, meeting the demands of global clients without compromising sustainability,” Costa added.

Hoping to serve technology giants and cloud providers, Tecto’s goal is to expand digital infrastructure supply in southern Brazil and strengthen Porto Alegre’s digital ecosystem, consolidating the city as a critical hub for connectivity and data infrastructure.

“This investment is a direct result of the position Rio Grande do Sul has achieved in recent years, driven by consistent efforts by the state government to transform the state into an international reference hub for technology and digital infrastructure, leading the country in innovation according to the Center for Public Leadership ranking,” said Rio Grande do Sul Governor Eduardo Leite.

“Connecting the new data centre to the Malbec submarine cable definitively places Rio Grande do Sul on the global data route, increases competitiveness, attracts high value-added businesses and generates sustainable economic development, with more qualified jobs, innovation and opportunities for the people of Rio Grande do Sul.”

Related stories

FIND OUT MORE: Capacity Latam 2026

AFD and IDB launch $324m subsea fibre and data centre project in Brazil’s Amazon

KIO CEO Octavio Camarena on AI readiness in LATAM

The agreement, revealed on Thursday, marks a major strategic shift for the Oslo-based operator as it continues to pull back from Asian markets to focus more squarely on its Nordic core.

Under the terms, Telenor will immediately sell 24.95% of its True shareholding at THB 11.70 per share to Arise Digital Technology Company, a firm controlled by Thai investor Khun Suphachai Chearavanont. A mutual put/call option covers the remaining 5.35%, scheduled for sale two years after the initial closing at the higher of the original price or prevailing market value.

True Corporation, created by the 2023 merger of True and dtac, is one of Thailand’s largest telecom and tech groups with around 60 million customers across mobile, broadband and digital services.

For Telenor, the sale caps a quarter-century in a market it first entered in 2000 and helped transform through major investments and the dtac–True merger. According to company figures, the value of Telenor’s Thai shareholding has grown from around NOK 12bn before merger talks to roughly NOK 39bn under the agreed sale terms.

The deal will also generate a significant accounting gain, with Telenor expecting to recognise about NOK 14.7bn once the first tranche closes. Further details on how the proceeds will be used are due alongside Telenor’s upcoming Q4 2025 results.

This divestment follows Telenor’s exit from Pakistan and underscores a broader pivot away from Asia. The transaction remains subject to customary regulatory approvals and is expected to complete in the coming months.

RELATED STORIES

Telenor expands fibre reach with $597m GlobalConnect deal

A guide to the Thai telecom tower market

True, Dtac complete largest Southeast Asia telco merger – Capacity